The use cases for this are unlimited. For example, imagine a simple general-purpose 'browser'-like client that took a txid as the URL. That client could then query any available archive node for that txid and display the contents of the 100k op_return as a webpage - with the op_return potentially including references to other txid's, and so on. BOOM: your content hosted for you forever for a one-time fee. Frankly, people have to be pretending that they lack the imagination to see the potential for the low cost of forever data amortized over eternity.